Overview

Autonomous agents are moving from lab prototypes to the core of enterprise operations. But turning powerful language models into reliable, observable, and governable systems is a hard engineering problem-one that spans architecture, tooling, security, and production operations. Autonomous AI Engineering with Microsoft is a deeply practical, architecture-first guide to building real-world, enterprise-grade agent systems on the Microsoft stack. Through the end-to-end design and implementation of a reference platform called AgentDesk, you'll learn how to go far beyond simple ""chatbots"" and prompts to deliver multi-agent, policy-aware, production-ready systems that enterprises can actually trust. This book is written for engineers, architects, and technical leaders who want to build serious autonomous systems-not just demos. You'll learn how to: Design robust agent architectures that separate concerns between orchestrators, coordinators, and specialist agents, while keeping governance and safety at the center. Implement core agents and toolchains using .NET, Python, Semantic Kernel, MCP, AutoGen, and Azure AI-connecting them to real enterprise systems like CRMs, ERPs, and internal APIs. Engineer memory and state with episodic, long-term, and vector-based semantic memory so agents can learn from experience instead of ""forgetting"" every interaction. Build multi-agent collaboration workflows where specialized agents coordinate, delegate, debate, and converge on decisions across complex business processes. Design prompts, policies, and guardrails that clearly define system intent, enforce hard constraints, and implement robust refusal logic and safety behavior. Integrate enterprise tooling and data through REST APIs, OpenAPI-based tool generation, webhooks, Azure AI Search, Cosmos DB, and SQL. Deploy and scale on Azure with Functions, App Service, and Container Apps; package agents in Docker; and wire up CI/CD pipelines using GitHub Actions and Azure DevOps. Instrument observability and reliability using metrics, structured logging, trace spans, Azure Monitor, and Application Insights-plus canary releases and automatic rollback. Optimize cost and performance with intelligent model selection, context window optimization, caching, and auto-scaling strategies tuned for LLM-heavy workloads.

Full Product Details

9798275284003

Table of Contents

Reviews

Author Information

Tab Content 6

Countries Available